Proceedings of DET2011 7th International Conference on Digital Enterprise Technology

Athens, Greece 28-30 September 2011

83

METROLOGY ENHANCED TOOLING FOR AEROSPACE (META): A LIVE FIXTURING, WING BOX ASSEMBLY CASE STUDY

O. C. Martin University of Bath

O.C.Martin@bath.ac.uk

Z. Wang University of Bath

zw215@bath.ac.uk

P. Helgosson University of Nottingham

epxph3@nottingham.ac.uk

J. E. Muelaner University of Bath

J.E.Muelaner@bath.ac.uk

A. Kayani Airbus UK

Amir.Kayani@airbus.com

D. Tomlinson Airbus UK

David.Tomlinson@Airbus.com

P. G. Maropoulos University of Bath

P.G.Maropoulos@bath.ac.uk

ABSTRACT

Aerospace manufacturers typically use monolithic steel fixtures to control the form of assemblies;

this tooling is very expensive to manufacture, has long lead times and has little ability to

accommodate product variation and design changes. Traditionally, the tool setting and

recertification process is manual and time consuming, monolithic structures are required in order to

maintain the tooling tolerances for multiple years without recertification. As part of a growing

requirement to speed up tool-setting procedures this report explores a coupon study of live

fixturing; that is, automated: fixture setting, correction and measurement. The study aims to use a

measurement instrument to control the position of an actuated tooling flag, the flag will

automatically move until the Key Characteristic (KC) of the part/assembly is within tolerance of its

nominal position. This paper updates developments with the Metrology Enhanced Tooling for

Aerospace (META) Framework which interfaces multiple metrology technologies with the tooling,

components, workers and automation. This will allow rapid or even real-time fixture re-certification

with improved product verification leading to a reduced risk of product non-conformance and

increased fixture utilization while facilitating flexible fixtures.

KEYWORDS

Dimensional Metrology, Measurement, Tooling, Fixture, Assembly, META

1. BACKGROUND

In the wider community tooling can include a wide

spectrum of tools, in the context of this paper

tooling is used to refer to Assembly Tooling, this

encompasses both Jigs and Fixtures

Monolithic aerospace assembly fixtures consist

of large traditional steel structures configured for a

single aircraft. For stability, the tooling is secured to

the reinforced-concrete factory floor. This

traditional build philosophy controls all the features

by: common jig location, master jig datum, jig

setting, certification points, build slips and pin

diameters. The positional, dimensional and

geometric accuracy of the assembly is implied from

the tooling. That is to say, if the tooling is correct

and the components are positioned correctly within

the tooling, then the assembly is correct. These

mechanical metrology checks ensure tolerances are

maintained.

84

Verification involves manually rotating pins and

moving slips to ensure that the assembly is correctly

positioned and held within the fixture. However, the

combined tolerance of the fixture and location

pins/slips must be less than the assembly tolerances;

ideally <10% although this is rarely possible at the

wing assembly scale; often the design tolerances are

<300µm over 30m. Subsequently, tooling is built to

a tolerance of around 150µm, consuming up to 50%

of the assembly tolerance budget. Next Generation

Composite Wings (NGCW) hold new challenges as

the composite materials cannot be reworked easily

if concessions are identified. Consequently, more

accurate assemblies - and therefore assembly

fixtures - will be required. These requirements will

further drive up the cost of traditional fixtures.

In addition, the size and complexity of fixtures

means that they typically have construction lead

times in excess of 6 months making late design

changes or the employment of concurrent

engineering a challenge. It is estimated that

assembly tooling accounts for approximately 5% of

the total build cost for an aircraft (Rooks, 2005) or

10% of the cost for the air frame (Burley et al,

1999). Fixture manufacture times and non-recurring

costs (NRCs) could be reduced if assembly fixtures

moved away from traditional hard tooling and

moved towards soft tooling, that is: away from

large, rigid structures and towards reconfigurable

and flexible tooling. A strong measurement

platform and infrastructure is required to maintain

the required tolerances within the tooling and the

assembly process.

2. INTRODUCTION

The key requirement for large-scale assembly is to

overcome the constraints associated with the

physical size of products and assemblies and the

corresponding dimensional and form tolerances

(Maropoulos et al, 2008). Advances in large volume

metrology are increasingly important in order to

achieve this; subsequently the realisation of

metrology enhanced tooling will become possible.

2.2. METROLOGY ENHANCED TOOLING FOR AEROSPACE (META)

Gauge-less and fixture-less manufacture are reliant

on the exploitation of advanced metrology in the

dimensional inspection and monitoring of the

tooling, components and assemblies. Firmly

embedded metrology systems within the

manufacturing processes are still not a reality as

most systems are outside of the tooling, and not

embedded within it. Metrology-assisted robotic

processes are being developed within manufacturing

cells with an emphasis on assembly, and not

conventional automated drilling processes

(Jayaweera and Webb, 2010). In order to place

metrology systems within the control loop of a

manufacturing cell prerequisites such as:

autonomous operation, high reliability, high speed

measurement, and flexibility are paramount (Gooch,

1998). This exploitation of technologies is stifled

due to the lack of integration with core design and

assembly processes.

The future of metrology enhanced tooling relies

on the effective synergy of complimentary

interfaces accommodated by a strong software

platform. These hybrid systems could utilise many

metrology technologies, for example: a macro co-

ordinate system could be set-up using

photogrammetry or a network of lasers – this would

effectively surround and monitor key characteristics

of the tooling. Localised metrology could sit within

this larger metrological environment – laser radar,

portable co-ordinate measurement machines

(PCMMs), actuators, sensors, arms, scanners, etc. –

providing fine measurement of difficult features,

freeform surfaces, tooling pick-ups, part location

and verification. Potentially this environment could

provide the prerequisite of any automation attempt,

determining the sources and magnitude of any

dimensional variations of the components that are

currently being experienced during the manual

assembly stage (Saadat and Cretin, 2002). Figure 1

gives an overview of the Metrology Enhanced

Tooling Aerospace (META) environment

introduced by Martin et al (2010).

The META framework’s primary function is to

monitor the key characteristics of the tooling and

assembly requiring a real-time or quasi-real-time

metrology system – ensure the fixture condition.

This monitoring eliminates the need to recertify

fixtures periodically removing the need to take the

fixture out of production – current practice can take

weeks to recertify and rework a fixture, causing

down time that will have increasing impact as

production rates increase. Secondary functions -

Enhanced Processes - such as ‘live’ tooling do not

require real-time feedback as the movements can be

iterative, unlike a machining operation. Machining

operations and automation where an iterative loop is

not appropriate must run directly from information

fed from the instrument – for example a laser

tracker – and not through the core software.

The META framework’s tertiary function is the

collection of information. This information could

not only enhance the tooling and assembly during

operation, but begin a large scale data collection for

the use of SPC, providing learning for future

optimization of the assembly processes.

85

Figure 1 - META Framework Overview

2.2. DATA FUSION

The META framework relies on instrument

networks for a number of reasons, mainly: reducing

measurement uncertainty, increasing the

measurement volume and providing complementary

technologies to enhance data collection. Due to the

expense of measurement instruments, instrument

networks can be performed by roving or multi-hop

systems using a single instrument many times.

Instrument hardware networks have many

challenges; using the data from each instrument in

the most efficient way is paramount. As different

instruments have differing strengths – data

management has to have an awareness of such

attributes and respond appropriately. Multi-sensor

data fusion is a method for centrally combining and

processing data from a number of different sensors

(Huang et al, 2007). The data fusion can be

described as either: complementary, competitive

and cooperative (Durrantwhyte, 1988).

Complementary if sensors are independent but can

offer additional information by complementing

other another; competitive if the sensors are

independently measuring the same area/targets in

order to eliminate random error and reduce

measurement uncertainty; and co-operative sensors

are independent but different from each other and

the combination of sensors provides a level of

information that each sensor cannot achieve alone.

Within dimensional metrology, examples of such

multi-sensor data fusion include: field of image

fusion, tactile and optical coordinate metrology,

coherent and incoherent optical measuring

techniques, computed tomography and scanning

probe microscopy (Weckenmann et al, 2009). It is

likely that the future of multi-sensor data fusion will

become increasingly important as higher levels of

integration with fast processing speeds become a

necessity for full field - large volume metrology and

automation.

3. CASE STUDY: FIXTURE AUTOMATION

This paper looks at the Secondary Function of the

META framework, metrology as an enabler of

‘live’ fixtures, as part of a growing requirement to

speed up tool-setting procedures; that is, automated:

fixture setting, correction and measurement. The

case study aims to use a measurement instrument to

control the position of an actuated tooling flag, the

flag will automatically move until the Key

Characteristic (KC) of the part/assembly is within

tolerance of its nominal position. This reduces the

measurement uncertainty stack-up associated with

constructing and employing tooling held tolerances;

effectively the tolerance budget is only occupied by

the instrument’s uncertainty and not the

manufacturing tolerances of the fixture. In the

META framework the measurement can focus on

assembly/component and not the fixture.

86

In the case of this study the actuated flag is a

Hexapod from Physik Instrumente (PI) and the KC

is the hinge line axis that runs through the hinge

bracket's bore. This trial was carried out

concurrently with the Airbus Tooling Hub activities,

based at the University of Nottingham.

3.1. HEXAPOD LOCATION

The methods used to build the trial fixture (Figure

2) cannot perfectly align the native co-ordinate

frame of the hexapod (Figure 3) to the jig co-

ordinate frame; aligning these frames accurately

would be a time consuming and laborious exercise.

A more robust and quicker method is to identify the

location and orientation of the hexapod's frame and

transform the relevant information into the jig co-

ordinate system when required; if calculations are

completed with an appropriate degree of accuracy

no loss of information will occur when changing

from frame-to-frame. This method allows the

hexapod to be approximately placed in its nominal

position without considering the hexapod's position

and orientation. This speeds up the tool setting

processes – making reconfiguration fixtures

quicker. However, in order to manipulate the

hexapod into its CAD nominal position, firstly the

hexapod’s native co-ordinate frame must be defined

relative to the jig axis system.

Figure 2 - Location of study on the demonstration fixture;

highlighted: the jig’s co-ordinate frame

Figure 3 - &ative hexapod co-ordinate frame in its CAD

nominal position

The hexapod is moved to the extreme of each

axis in isolation using PI's proprietary software

interface (Figure 4); each axis extremity is

measured using a Leica AT901 laser tracker and

!ew River Kinematic’s: SpatialAnalyzer (SA). This

enables the definition of the working envelope

(x=50mm, y=50mm, z=25mm) and the creation of

the physical, native co-ordinate frame of the

hexapod relative to the fixture's co-ordinate frame.

Subsequently, the hexapod can be manoeuvred into

its CAD nominal position by obtaining the

translations [x, y, z]T and rotations [α, β, γ]

T from

the SA function: compare to CAD. This method is

consistent with the fixture build philosophy used for

the construction of the fixture. In turn the physical

location of hexapod’s frame can be compared to the

CAD nominal location of the hexapod frame

(Figure 5). The transformation matrix from native to

CAD nominal (Equation-1) gives us the offsets

required to reach the intended CAD nominal

position.

This is a specific transformation matrix that uses

the sequence: rotate about x (α), followed by y

rotation (β), then rotated about z (γ), finally,

performing a translation in x, y, z; this is the

sequence that the SA software uses.

−

−+

+−

=

1000

coscossincossin

sincoscossinsincoscossinsinsincossin

sinsincossincoscossinsinsincoscoscos

t

t

t

z

y

x

Tγβγββ

γαγβαγαγβαβαγαγβαγαγβαβα

(1)

87

Figure 4 - PI hexapod controller interface

Figure 5 - Actual position of the Hexapod's native co-

ordinate frame

3.2. HEXAPOD COMMUNICATION AND CONTROL

The measurement information from the laser tracker

is continuously streamed into SA. SA converts the

native spherical co-ordinates from the laser tracker

to the Cartesian co-ordinates required for the

hexapod control. This post-processed data is

streamed via a User Datagram Protocol (UDP) to a

bespoke program created by the University of Bath

(UoB); designed to bridge the interface gap between

the PI hexapod interface and SA. The UoB interface

program (Figure 6) samples the UDP data stream,

checks whether the KC is within tolerance, sends

the required corrective movement to the hexapod

and checks whether the hexapod is stationary before

cycling again. The communication paths between

the hardware and software are shown in Figure 7.

The program also enabled the control of a selection

of parameters, such as: the tolerance threshold,

hexapod velocity, enabling and disabling the

hexapod's degrees of freedom and closed or open

loop control.

Figure 6 - UoB SA - Hexapod interface software

88

Figure 7 - Schematic of hardware/software communication

3.3. METROLOGICAL FEEDBACK

The metrology requirement is to measure the

deviation from the hinge bracket's bore to its

nominal CAD position; the hexapod will move,

attempting to reclaim the hinge line’s CAD nominal

position. The hexapod is attached to the spar via a

zero point clamp (Figure 8), there is a substantial

offset from the point of attachment to the point of

interest (POI) (Figure 9) between hexapod and the

POI are compliant connective elements: zero point

clamp, spar, hinge bracket and vector bar. As the

relationship between the hexapod and the POI

cannot be considered as a rigid body the metrology

feedback will have to be in a closed loop (Figure

10); if however there was a rigid relationship or a

predicable relationship between the movement of

the hexapod and the POI, the PI hexapod is

inherently accurate enough to support an open loop

system - this is quicker and less resource intensive

(Figure 11). An open loop system is advantageous

when considering measurement resources and time;

a closed loop system requires continuous

measurement, whereas an open loop system requires

a single measurement. If many POIs require

measurement and actuation, closed loop systems are

bottlenecked by the metrology resource, this

happens to a much lesser extent with an open loop

system; as the measurement system can sequentially

measure each POI without stopping.

Figure 8 - Close up of the Zero point clamp

Figure 9 highlights the hexapod's native co-

ordinate after the origin has been translated to the

POI; it follows that measurements taken from this

new co-ordinate frame are essentially deviations

from the POIs nominal position. Consequently, the

co-ordinates - and hence the deviations from

nominal - are streamed from SA via the UDP.

Figure 9 - Hexapod's native co-ordinate frame after

transformation to CAD nominal position of hinge line axis

89

Figure 10 - Closed loop control of fixture

Figure 11 - Open loop control of fixture

The measurement instrument used for the trial

was a Leica AT901 laser tracker, this was a readily

available instrument with a good level of accuracy

capable of real-time, three dimensional

measurement. 3D co-ordinates were assumed as

appropriate since the compliance of the material is

limited to two dimensions and this phase of the

trials is assessing the feasibility of the metrology

enhanced tooling philosophy. Figure 12 shows the

laser tracker point of measurement relative to the

zero-point clamp attachment point.

Figure 12 - Facility tooling for targeting the hinge line

4. RESULTS

The trial focus on moving two axis, without

rotation; engaging the hexapod's y-axis (Figure 9:

Green Arrow) and z-axis (Figure 9: Blue Arrow).

The reason for not actuating the x-axis was

structural: as longitudinal movement was likely to

add additional stress to the fasteners as the structure

had high rigidity in this plane. Rotational

movements were excluded at this stage because

only one POI was monitored, rotational movement

is more appropriate when best-fitting multiple

points. The most out of tolerance axis was z. The

closed loop configuration moved the POI a total of

0.421mm in the y-axis and negative 1.572mm in z-

axis; the hexapod achieved the designated tolerance

threshold (±300µm) within two iterative cycles –

this is summarized in Figure 13.

However, the hexapod's encoders registered a

movement of 1.103mm in the y-axis (Figure 14) and

negative 2.412mm in the z-axis (Figure 15); this

90

difference can be attributed to the material

compliance. This confirms the assumption that the

POI and hexapod do not act as a rigid body.

However, Figure 14 and Figure 15 show that

after around 5 iterations the deviation between the

POI and hexapod displacements begin to level out,

reducing the significance of the component

deflection and offset.

Figure 13 - POI displacement from nominal in y-axis and z-axis after each move iteration, with measurement uncertainty bars

indicated

Figure 14 - Measured POI displacement (with uncertainty indicated) compared with displacement from hexapod's encoders; in

the y-axis

91

Figure 15 - Measured POI displacement (with uncertainty indicated) compared with displacement from hexapod's encoders; in

the z-axis

5. CONCLUSIONS

Figure 16 shows the elements of the META

framework exercised through the live fixturing

study.

The closed loop model holds obvious limitations

in terms of measurement resources; if the fixturing

requires 3DOF or 6DOF manipulation within a

global co-ordinate system then the measurement

instrument is likely to be prohibitively expensive for

deployment on each actuated part. Subsequently,

the closed loop model has to multi-hop instruments

to each actuated pick-up. This is an inherently time

consuming process; however, bottle-necking due to

metrology resource on multiple POIs could be

reduced by cycling through each of the POIs and

assuming that a small number of iterations is

necessary to achieve the tolerance. This would

negate the requirement of metrology monitoring and

is substantiated in Figure 13, Figure 14 and Figure

15. However this is reliant on a degree of actuator

accuracy.

On the other hand, a closed loop model does

mean that the actuators do not need to be accurate,

just a fine resolution of movement. If the pick-up

only needs local measurement or describing in one

or two dimensions then inexpensive measurement

systems could be deployed on each manipulator and

closed loop systems could be used.

Open loop systems may be a more economical

solution as an enabler to live fixturing; one laser

tracker or photogrammetric survey could measure

all the POIs and the accuracy on the actuators could

be relied on to position the pick-ups to within

tolerance. However, this would need rigid body

relationships to be established, or known deflections

compensated for.

92

Figure 16 - META Framework with aspect utilized in hexapod control highlighted

6. ACKNOWLEDGMENTS

We would like to thank Steven Lockett and Geraint

Green (PI) for your technical support; as well as Dr

Tony Smith and the team (University of

!ottingham) for co-ordinating resource and hosting

the project.

This paper is part of a PhD sponsored by Airbus

UK and the EPSRC Innovative Manufacturing

Research Centre at the University of Bath (grant

reference GR/R67507/0).

REFERENCES

Burley G, Odi R, Naing S, Williamson A and Corbett J,

“Jigless aerospace manufacture - The enabling

technologies”, in Aerospace Manufacturing

Technology Conference & Exposition. Bellevue,

Washington: Society of Automotive Engineers, 1999

Durrantwhyte HF, “Sensor models and multisensor

integration”, International Journal of Robotics

Research, Vol. 7, No. 6, 1988, pp 97-113

Gooch R, “Optical metrology in manufacturing

automation”, Sensor Review, Vol. 18, No. 2, 1998, pp

81-87

Huang Y, Lan Y, Hoffmann WC and Lacey RE,

“Multisensor Data Fusion for High Quality Data

Analysis and Processing in Measurement and

Instrumentation”, Journal of Bionic Engineering, Vol

4, No. 1, 2007, pp 53-62

Jayaweera N and Webb P, “Metrology-assisted robotic

processing of aerospace applications”, International

Journal of Computer Integrated Manufacturing, Vol.

23, No. 3, 2010, pp 283-296

Maropoulos PG, Guo Y, Jamshidi J and Cai B, “Large

volume metrology process models: A framework for

integrating measurement with assembly planning”,

CIRP Annals - Manufacturing Technology, Vol. 57,

No. 1, 2008, pp 477-480

Martin OC, Muelaner JE, Tomlinson D, Kayani A and

Maropoulos PG, “Metrology Enhanced Tooling for

Aerospace (META) Framework”, Proceedings of the

36th

International MATADOR Conference,

Manchester, 2010, pp 363-366

Rooks B, “Assembly in aerospace features at IEE

seminar”, Assembly Automation, Vol. 25, No. 2, 2005,

pp 108-111

Saadat M and Cretin C, “Dimensional variations during

Airbus wing assembly” Assembly Automation, Vol.

22, No. 3, 2002, pp 270-279

Weckenmann A, Jiang X, Sommer KD, Neuschaefer-

Rube U, Seewig J, Shaw L and Estler T, “Multisensor

data fusion in dimensional metrology”, CIRP Annals -

Manufacturing Technology, Vol. 58, No. 2, 2009, pp

701-721